Delving into adversarial attacks on deep policies

TL;DR

This paper investigates adversarial attacks on deep reinforcement learning policies, comparing attack methods, proposing a new attack reduction technique, and examining the impact of retraining on robustness.

Contribution

It introduces a novel attack reduction method based on the value function and analyzes the effects of retraining on model resilience.

Findings

Adversarial examples are more effective than random noise in attacking policies.

The proposed method reduces the number of adversarial injections needed for success.

Retraining on noise and FGSM perturbations influences policy robustness.

Abstract

Adversarial examples have been shown to exist for a variety of deep learning architectures. Deep reinforcement learning has shown promising results on training agent policies directly on raw inputs such as image pixels. In this paper we present a novel study into adversarial attacks on deep reinforcement learning polices. We compare the effectiveness of the attacks using adversarial examples vs. random noise. We present a novel method for reducing the number of times adversarial examples need to be injected for a successful attack, based on the value function. We further explore how re-training on random noise and FGSM perturbations affects the resilience against adversarial examples.